Knowledge Graph entity type prediction using Graph Neural Networks and LLMs.
Master Research project at Télécom Paris (Team Data, Intelligence and Graphs). Supervised by Mehwish Alam.
The entity type information in Knowledge Graphs (KGs) such as DBpedia, Freebase, etc. is often incomplete due to automated generation or human curation. Entity typing is the task of assigning or inferring the semantic type of an entity in a KG. This PRIM projects aims at leveraging the outstanding semantic knowledge and understanding of new open source large language models (LLMs) such as Llama 3 8-b jointly with Graph Neural networks to perform automatic entity type prediction from given node and surrounding relationships with other entities in a Knowlege Graph.
An example of a DBpedia Knowledge Graph, with entities and their relationships.
